Tech and AI Acronyms Explained: The Complete Glossary

Why tech acronyms show up in business networking

Your group includes SaaS founders, CTOs, agency owners, and advisors serving tech clients. They say “we need RAG over our CRM” or “our MCP integration ships next sprint”—and expect you to follow.

You do not need to be an engineer. You need a reference. Use this list before meetings, when writing referral messages to tech buyers, or when a published need mentions AI, cloud, or security requirements.

AI, machine learning, and models

Core artificial intelligence terms—the foundation of most AI product conversations today.

• AI — Artificial Intelligence — Systems that perform tasks requiring human-like reasoning, perception, or language
• ML — Machine Learning — AI that learns patterns from data instead of explicit hand-coded rules
• DL — Deep Learning — ML using multi-layer neural networks
• NN — Neural Network — Computing model inspired by biological neurons; backbone of modern AI
• LLM — Large Language Model — AI model trained on vast text to generate and understand language (e.g., chat assistants)
• SLM — Small Language Model — Compact language model for edge devices or lower cost/latency
• GPT — Generative Pre-trained Transformer — Architecture family behind many language models; also used generically for chat AI
• Transformer — Neural architecture using attention mechanisms; standard for modern LLMs
• AGI — Artificial General Intelligence — Hypothetical AI matching human-level ability across all cognitive tasks
• ANI — Artificial Narrow Intelligence — AI specialized for one task (all current production AI)
• GenAI — Generative AI — AI that creates new content: text, images, code, audio, video
• Inference — Running a trained model to produce outputs (as opposed to training)
• Training — Process of teaching a model from data; compute-heavy
• Fine-tuning — Adapting a pre-trained model to a specific domain or task with additional data
• SFT — Supervised Fine-Tuning — Fine-tuning on labeled input-output examples
• RLHF — Reinforcement Learning from Human Feedback — Aligning models using human preference rankings
• DPO — Direct Preference Optimization — Alternative to RLHF for preference alignment
• PEFT — Parameter-Efficient Fine-Tuning — Fine-tuning methods that update only a small subset of weights (e.g., LoRA)
• LoRA — Low-Rank Adaptation — Popular PEFT technique; adds small trainable layers
• MoE — Mixture of Experts — Model architecture routing inputs to specialized sub-networks
• RL — Reinforcement Learning — Learning through rewards and penalties from outcomes
• HITL — Human In The Loop — Human review or correction in an automated AI workflow
• Token — Smallest unit models process (word pieces); usage often billed per token
• Context window — Maximum tokens a model can consider in one request
• Hallucination — Confident but incorrect AI output not grounded in facts
• Grounding — Tying model answers to verified sources or data
• Prompt — Input instructions and context sent to a language model
• CoT — Chain of Thought — Prompting technique asking the model to reason step-by-step
• Temperature — Setting controlling randomness/creativity of model outputs
• Embedding — Numeric vector representation of text, images, or data for similarity search
• Multimodal — Models handling more than one input type (text + image + audio)

MCP, agents, and AI integrations

How AI connects to tools, APIs, and workflows—the fastest-moving area in 2025–2026 product discussions.

• MCP — Model Context Protocol — Open standard for connecting AI models to external tools, data sources, and services in a consistent way
• Agent — AI system that plans and executes multi-step tasks using tools, not just single chat replies
• AI agent — Autonomous or semi-autonomous software using LLMs plus tools (search, code, CRM, email)
• Tool use / function calling — LLM invoking structured APIs (calendar, database, calculator) during a conversation
• RAG — Retrieval-Augmented Generation — Fetching relevant documents from a knowledge base before generating an answer
• Vector DB — Vector database — Storage optimized for embedding similarity search (Pinecone, Weaviate, pgvector, etc.)
• Knowledge base — Curated documents and data an AI can retrieve from
• Orchestration — Coordinating multiple models, agents, or steps in a pipeline
• Workflow automation — Chaining triggers and actions (often overlaps with agents and integrations)
• API — Application Programming Interface — Defined way for software systems to communicate
• SDK — Software Development Kit — Libraries and tools for building on a platform
• Webhook — HTTP callback triggered by an event in another system
• Plugin — Extension adding capabilities to a host application (IDE, chat, CRM)
• Copilot — AI assistant embedded in a product assisting the user in context
• A2A — Agent-to-Agent — Systems where multiple AI agents coordinate (emerging term; context varies)
• JSON — JavaScript Object Notation — Standard lightweight data format for APIs
• Schema — Structured definition of data fields models or APIs expect

NLP, speech, and vision

Language and perception AI—common in martech, support, and document-processing referrals.

• NLP — Natural Language Processing — AI understanding and generating human language
• NLU — Natural Language Understanding — Parsing intent and meaning from text
• NLG — Natural Language Generation — Producing human-readable text from data
• CV — Computer Vision — AI interpreting images and video
• OCR — Optical Character Recognition — Extracting text from images or scanned documents
• ASR — Automatic Speech Recognition — Speech-to-text
• TTS — Text-to-Speech — Synthetic voice from text
• STT — Speech-to-Text — Same domain as ASR
• NER — Named Entity Recognition — Finding names, places, organizations in text
• Sentiment analysis — Classifying text as positive, negative, or neutral
• Translation MT — Machine Translation — Automated language translation
• GAN — Generative Adversarial Network — Two-network architecture for generating realistic data
• Diffusion model — Generative image/video approach used by many modern media AI tools

Cloud, infrastructure, and DevOps

Terms SaaS founders and engineering leaders use when describing stack, scale, and reliability.

• SaaS — Software as a Service — Cloud software sold by subscription
• PaaS — Platform as a Service — Cloud platform to build and deploy apps (e.g., managed runtimes)
• IaaS — Infrastructure as a Service — Raw compute, storage, and networking in the cloud
• AWS — Amazon Web Services — Major cloud provider
• GCP — Google Cloud Platform — Google’s cloud services
• Azure — Microsoft’s cloud platform
• VM — Virtual Machine — Emulated computer on shared physical hardware
• Container — Lightweight packaged application with dependencies (Docker standard)
• Docker — Platform for building and running containers
• K8s — Kubernetes — System for orchestrating containers at scale
• Orchestration — Automated deployment, scaling, and management of containers or services
• Serverless — Cloud model where provider runs your code on demand without you managing servers
• Lambda — AWS serverless function product (often used generically for functions-as-a-service)
• EC2 — Elastic Compute Cloud — AWS virtual servers
• S3 — Simple Storage Service — AWS object storage
• CDN — Content Delivery Network — Distributed cache for faster global content delivery
• DNS — Domain Name System — Maps domain names to IP addresses
• Load balancer — Distributes traffic across multiple servers
• Auto-scaling — Automatically adding/removing capacity based on demand
• Uptime — Percentage of time a service is available
• SLA — Service Level Agreement — Contracted availability or performance guarantee
• Latency — Delay before a response; critical for AI and real-time apps
• Throughput — Amount of work processed per unit of time
• Cold start — Delay when a serverless function or model wakes from idle
• GPU — Graphics Processing Unit — Hardware accelerating AI training and inference
• TPU — Tensor Processing Unit — Google’s AI accelerator chips
• NPU — Neural Processing Unit — On-device AI accelerator in phones and PCs
• VRAM — Video RAM — GPU memory; limits model size on local hardware
• CUDA — NVIDIA platform for GPU computing
• On-prem — On-premises — Software running on company-owned servers, not cloud
• Hybrid cloud — Mix of on-prem and public cloud
• Multi-cloud — Using more than one cloud provider
• Region / AZ — Availability Zone — Isolated data center grouping within a cloud region
• IaC — Infrastructure as Code — Defining infrastructure in version-controlled files (Terraform, etc.)
• Terraform — Popular IaC tool for provisioning cloud resources
• CI — Continuous Integration — Automated build and test on each code change
• CD — Continuous Delivery/Deployment — Automated release to staging or production
• CI/CD — Combined pipeline from commit to deploy
• DevOps — Culture and practice merging development and operations
• GitOps — Operations model using Git as source of truth for deployments
• Monorepo — Single repository containing multiple projects or services
• Microservices — Architecture of small independent services vs one monolithic app
• Monolith — Single deployable application containing all features
• Legacy system — Older software still in production; hard to replace

Software development

Engineering vocabulary that appears in product specs, RFPs, and technical referral requests.

• IDE — Integrated Development Environment — Code editor with debugging and tooling (VS Code, JetBrains)
• CLI — Command Line Interface — Text-based tool control
• GUI — Graphical User Interface — Visual interface with buttons and windows
• OOP — Object-Oriented Programming — Code organized around objects and classes
• FP — Functional Programming — Code built from pure functions and immutability
• CRUD — Create, Read, Update, Delete — Basic data operations
• REST — Representational State Transfer — Common style for HTTP APIs
• GraphQL — Query language for APIs letting clients request exact fields needed
• gRPC — High-performance RPC framework, common for internal microservices
• HTTP — Hypertext Transfer Protocol — Foundation of web requests
• HTTPS — HTTP Secure — HTTP encrypted with TLS
• WebSocket — Persistent two-way connection for real-time apps
• ORM — Object-Relational Mapping — Library mapping database rows to code objects
• SQL — Structured Query Language — Standard language for relational databases
• NoSQL — Non-relational databases (document, key-value, graph, etc.)
• PostgreSQL / Postgres — Popular open-source relational database
• Redis — In-memory data store used for cache and queues
• MongoDB — Popular document NoSQL database
• Kafka — Distributed event streaming platform
• Message queue — System buffering tasks between services (RabbitMQ, SQS, etc.)
• ETL — Extract, Transform, Load — Moving and reshaping data between systems
• ELT — Extract, Load, Transform — Variant loading raw data before transformation
• Open source / OSS — Software with publicly available source code
• FOSS — Free and Open Source Software
• PR — Pull Request — Proposed code change for review before merge
• MR — Merge Request — Same concept as PR (GitLab terminology)
• SemVer — Semantic Versioning — Version numbers as MAJOR.MINOR.PATCH
• Breaking change — Update incompatible with previous API or behavior
• Deprecation — Feature marked for future removal
• RFC — Request for Comments — Formal proposal document (Internet standards or internal design docs)
• MVP — Minimum Viable Product — Earliest product version testing core value
• POC — Proof of Concept — Small experiment validating feasibility
• Spike — Time-boxed research task in agile development
• Tech debt — Shortcuts in code that speed now but cost later to fix
• Refactor — Restructuring code without changing external behavior
• Legacy code — Old codebase still maintained but difficult to extend

Security and compliance

Essential when referring to CISOs, regulated industries, or enterprise procurement.

• CISO — Chief Information Security Officer — Executive owning cybersecurity strategy
• InfoSec — Information Security — Protecting data and systems from unauthorized access
• AppSec — Application Security — Security of software applications
• DevSecOps — Integrating security into DevOps pipelines
• CVE — Common Vulnerabilities and Exposures — Public catalog of known security flaws
• XSS — Cross-Site Scripting — Attack injecting malicious scripts into web pages
• CSRF — Cross-Site Request Forgery — Forcing authenticated users to execute unwanted actions
• SQLi — SQL Injection — Attack manipulating database queries through input
• MFA — Multi-Factor Authentication — Login requiring two or more verification factors
• 2FA — Two-Factor Authentication — MFA with exactly two factors
• SSO — Single Sign-On — One login for multiple applications
• OAuth — Open Authorization — Standard for delegated access without sharing passwords
• OIDC — OpenID Connect — Identity layer on top of OAuth
• JWT — JSON Web Token — Compact token format for auth and API sessions
• SAML — Security Assertion Markup Language — Enterprise SSO standard
• RBAC — Role-Based Access Control — Permissions assigned by role
• IAM — Identity and Access Management — Policies for who can access what
• PKI — Public Key Infrastructure — System managing digital certificates
• TLS — Transport Layer Security — Encryption protocol for HTTPS
• SSL — Secure Sockets Layer — Predecessor name still used colloquially for TLS
• VPN — Virtual Private Network — Encrypted tunnel for remote network access
• Zero Trust — Security model assuming no implicit trust inside the network
• SOC — Security Operations Center — Team monitoring threats (also Service Organization Control for audits)
• SOC 2 — Audit standard for service providers on security controls
• SIEM — Security Information and Event Management — Centralized security log analysis
• DDoS — Distributed Denial of Service — Overwhelming a service with traffic
• WAF — Web Application Firewall — Filters malicious HTTP traffic
• GDPR — General Data Protection Regulation — EU privacy law
• HIPAA — US health data privacy and security law
• PII — Personally Identifiable Information — Data identifying an individual
• PHI — Protected Health Information — Health-related PII under HIPAA
• Pen test — Penetration test — Simulated attack to find vulnerabilities
• Red team / Blue team — Offensive vs defensive security exercises

Data, analytics, and business intelligence

When members refer data engineers, analytics leads, or BI projects.

• BI — Business Intelligence — Reporting and dashboards for business decisions
• DW — Data Warehouse — Central repository for analytical data
• Data lake — Storage holding raw structured and unstructured data
• Data mesh — Decentralized architecture treating data as a product by domain
• CDC — Change Data Capture — Tracking database changes for sync or analytics
• OLTP — Online Transaction Processing — Databases optimized for day-to-day operations
• OLAP — Online Analytical Processing — Databases optimized for analytics queries
• dbt — Data build tool — SQL transformation workflow for analytics teams
• Spark — Apache Spark — Large-scale data processing engine
• Hadoop — Ecosystem for distributed storage and processing of big data
• Parquet — Columnar file format common in data lakes
• CSV — Comma-Separated Values — Simple tabular data format
• YAML — Human-readable data format for config files
• XML — Extensible Markup Language — Structured document format (legacy integrations)
• Dashboard — Visual summary of KPIs and metrics
• KPI — Key Performance Indicator — Metric tracking progress to a goal
• Metric / dimension — Measure vs category used to slice data in analytics
• A/B test — Experiment comparing two variants to measure impact
• Cohort analysis — Tracking behavior of user groups over time
• Data pipeline — Automated flow moving and transforming data between systems

Product, UX, and delivery

Product and design terms in SaaS and agency referral conversations.

• UX — User Experience — How people feel using a product
• UI — User Interface — Visual layout and controls users interact with
• DX — Developer Experience — Quality of experience for developers using your API or platform
• PM — Product Manager — Owns problem definition, roadmap, and priorities
• PD — Product Designer — Designs UX/UI (title varies)
• QA — Quality Assurance — Testing before release
• UAT — User Acceptance Testing — Final validation by business users
• Sprint — Fixed time-box in agile development (often two weeks)
• Scrum — Agile framework with sprints, standups, and retrospectives
• Kanban — Visual workflow method limiting work in progress
• Agile — Iterative delivery approach vs big-bulk waterfall releases
• Waterfall — Sequential project phases; common in legacy enterprise
• Backlog — Prioritized list of work not yet done
• Roadmap — Planned features and milestones over time
• Wireframe — Low-fidelity layout sketch
• Prototype — Interactive mockup for testing before build
• MVP — Minimum Viable Product — Smallest release to learn from real users
• PMF — Product-Market Fit — Strong demand signal for the product
• Low-code — Building apps with minimal hand-coding via visual tools
• No-code — Building apps without traditional programming
• RPA — Robotic Process Automation — Bots automating repetitive UI tasks
• B2B2C — Business selling through partners to end consumers
• White-label — Product rebranded and sold by another company
• API-first — Product designed around programmatic access from the start

Web, mobile, and connectivity

Basics for referrals involving digital agencies, e-commerce, or app builders.

• SEO — Search Engine Optimization — Improving organic search visibility
• SEM — Search Engine Marketing — Paid search advertising
• CMS — Content Management System — Software managing website content (WordPress, etc.)
• SSR — Server-Side Rendering — HTML generated on server per request
• SPA — Single Page Application — Web app loading once then updating dynamically
• PWA — Progressive Web App — Web app with app-like features offline and on home screen
• API rate limit — Cap on requests per time window
• Mobile native — App built specifically for iOS or Android
• Cross-platform — One codebase targeting multiple platforms (React Native, Flutter)
• iOS — Apple mobile operating system
• Android — Google mobile operating system
• IoT — Internet of Things — Connected physical devices sending data
• Edge computing — Processing data near the source instead of only in central cloud
• 5G — Fifth-generation cellular network standard
• Bandwidth — Data transfer capacity of a connection
• IP — Internet Protocol — Addressing and routing packets on networks
• IPv4 / IPv6 — Older and newer IP address standards
• FTP / SFTP — File transfer protocols (SFTP is secure)

Emerging and blockchain (select terms)

Less common in every meeting but appears in fintech, gaming, and Web3 referrals.

• Web3 — Vision of decentralized internet often built on blockchains
• Blockchain — Distributed ledger recording transactions across nodes
• Smart contract — Self-executing code on a blockchain when conditions are met
• NFT — Non-Fungible Token — Unique digital asset recorded on chain
• DeFi — Decentralized Finance — Financial services on blockchain without traditional banks
• DAO — Decentralized Autonomous Organization — Community-governed entity via tokens and votes
• Crypto — Cryptocurrency — Digital assets using cryptography (context: currency vs cryptography)
• Wallet — Software storing keys to access blockchain assets
• KYC — Know Your Customer — Identity verification required by regulated finance
• AML — Anti-Money Laundering — Compliance controls against illicit funds

How to use this list when networking

When a member publishes a need involving AI or tech, ask which acronyms in their stack matter for a fit check—MCP integration, SOC 2, Salesforce API, etc.

When referring to a technical buyer, mirror their vocabulary accurately in your referral message. Misusing LLM, RAG, or MCP signals low fit.

Bookmark this page and the business acronyms glossary together—they cover most B2B referral conversations in mixed professional and tech groups.